Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!sun-barr!cs.utexas.edu!usc!hacgate!tcville!sed170!lee
From: lee@sed170.HAC.COM (John Lee )
Newsgroups: comp.sys.amiga.tech
Subject: Re: Chroma-key (was: Chaining Genlocks)
Message-ID: <304@sed170.HAC.COM>
Date: 26 Jul 89 17:48:04 GMT
References: <367@cbnewsd.ATT.COM> <7245@cbmvax.UUCP> <15742@pasteur.Berkeley.EDU>
Reply-To: lee@sed170.UUCP (John Lee (ird))
Followup-To: comp.sys.amiga.tech
Organization: Hughes Aircraft Co., El Segundo, CA
Lines: 52

In article <15742@pasteur.Berkeley.EDU> johnhlee@cory.Berkeley.EDU.UUCP (John Lee) writes:
[...hypothesis that genlocks are switched via color0 signal from Amiga...]
> What I was thinking was this:  I remember that the old commodore genlock
> for the 1000 converted the incoming video to RGB and showed the genlocked
> image on the RGB passthru (since this is just an analog-to-analog conversion,
> this doesn't surprise me).  I don't know if any of the genlocks on the 
> market (or which) do this too.  The ones i've seen only have composite out.
> Anyway, if you built a little circuit which looked at the RGB signals on this
> line and outputted a digital signal depending on how "blue" the output was,
> you could feed this signal back into the genlock instead of the color0 line
> and have a primitive and CHEAP chroma-key.  True, you couldn't have any
> blue hues in your computer graphics, but whadda you want for (almost) free?
>
> Any replies?  would it be fast enough?  Please don't chastise me too much
> if i've said something dumb, though.
>
> - Vince Lee (now back from Europe), using my brother's account since
>   berkeley doesn't think Mech E's deserve computer time.

Good idea, in concept, but you'll need to do a little circuit hacking in order
for that to work.  First of all, the composite-to-RGB conversion must take
place before the video switcher in the genlock, and the "blueness" circuit
must take its input from there.  Otherwise, the "blueness" circuit detects
a suitable blue, commands the genlock to switch to non-blue computer graphics,
which turns off the external video, thus turning off the "blueness" circuit,
which commands the genlock to switch back to external blue video, which turns
back on the "blueness" circuit...

See a problem here?  In other words, the RGB out can't be derived from the
switched video output from the same genlock.  But use two chained genlocks
with the first used only for its composite-to-RGB converter, and you *might*
have something workable.

Then there's the problem with delay.  Hi-res pixels are only 70ns long, and
a simple TTL NAND gate has about 8ns propagation delay.  The circuit must
be carefully designed so that the blue-detect and switch circuits have
minimum delay, otherwise colorfringing and other unpleasant color-interference
problems could be present.  A good, clean, accurate analog delay (does such a
thing exist?) on the video-in just before the switcher would eliminate this
problem.

--John Lee

Gee, am I talking to myself?  I don't think Mech E's deserve computer time,
either.  :-)
-------------------------------------------------------------------------------
Raining CATS and DOGS?  Join the RATS: Remote Amiga Teleconferencing System
	+--------+			John Lee
	| HUGHES |
	+--------+			ARPAnet: jhlee@hac2arpa.hac.com	
	Hughes Aircraft Company
The above opinions are those of the user and not of those of this machine.